{"id":"ad6039e6-3321-44a5-a18b-8ad8764395ee","arxiv_id":"2606.21826","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"DESI data hint that dark energy evolves (w0>−1, wa<0, phantom crossing) at 2.5–4.2σ, but the same expansion history can also be produced by interacting dark sectors, modified gravity, or non-cold dark matter.","lead":"This review collects what DESI baryon-acoustic-oscillation data, combined with supernovae and the cosmic microwave background, currently say about dark energy: a 2.5–4.2σ hint that its equation of state evolves and crosses the phantom divide. It argues the signal is real but not proven, because the same expansion history can be mimicked by interacting dark energy, modified gravity, or non-cold dark matter.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The DDE 'indication' is not yet robust: §2.3 shows it depends on the DESY5 low-z anchor and the DESI LRG2 bin, with recalibration lowering significance from ~4.2σ to ~3.2σ. The review's conclusion treats this as a real departure from ΛCDM—the least secure step.","rationale":"The review's argument is a synthesis, so I ask what must be true for its central claim to hold: the quoted w0>−1, wa<0 preference must reflect the actual distance–redshift relation rather than a few-percent calibration effect. The paper's own §2.3 identifies two concrete failure modes (DESY5 low-z anchor, LRG2 bin) and cites studies showing that correcting either weakens the signal. This makes the empirical premise the least secure link in the chain. It is not an outside-consensus objection; it is documented in the text and in the cited literature. I credit the review for explicitly flagging these limitations and for repeatedly calling the result an 'indication' rather than a discovery; that hedging is genuine. But hedging does not make the premise load-bearing—it makes the conclusion conditional. The reader's verdict is already CONDITIONAL and identifies the same weakest assumption, so no verdict change is needed. I set aside the secondary concerns about self-citation and the deferred 'Zhang 2026' review: those affect completeness and balance, not the empirical core. The proposed concrete test isolates exactly the two systematics the review treats as most influential; if the preference survives that joint reanalysis, the concern is settled rather than merely asserted.","tokens_in":48230,"tokens_out":10702,"duration_ms":101040,"concrete_test":"One decisive check: re-fit the CPL model to DESI DR2 BAO + Planck CMB + DESY5 using (i) the DES-Dovekie photometric recalibration in place of the original DESY5 high/low-z anchor and (ii) a DESI BAO likelihood with the z_eff≈0.51 LRG bin removed, holding all priors and covariance choices identical to the cited analyses. Report Δχ² (or ΔlnZ) relative to w=−1, wa=0 and the 1σ/2σ contours in the (w0, wa) plane. If the preference falls below 2σ or the 1σ contour includes ΛCDM, the review's central 'indication' is not robust to the two systematics it itself identifies; if it remains ≥3σ, the signal survives the strongest documented calibration/driver concerns.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that DESI BAO + CMB + SN data currently indicate a departure from ΛCDM characterized by w0 > −1, wa < 0, motivating a menu of beyond-ΛCDM interpretations. The load-bearing premise is that this inferred trend is a property of the distance–redshift relation, not a calibration artifact. The review itself, in §2.3, supplies the main evidence that this premise is insecure: a few hundredths-of-a-magnitude offset between the DESY5 low- and high-z samples can mimic an evolving EoS (Efstathiou 2025; Huang et al. 2025); the DES-Dovekie recalibration reduces the DR2 significance from ~4.2σ to ~3.2σ (Popovic et al. 2026); the DESI LRG bin at z_eff≈0.51 implies Ωm≈0.67 and drives the DDE shift (Colgáin & Sheikh-Jabbari 2025); and removing the LRG2 D_V point moves the (w0, wa) contour toward consistency with w=−1 within 2σ (Wang 2024). These are exactly the conditions under which 'current analyses report a preference' and 'there is a real deviation' come apart. If the trend disappears under a conservative treatment of these two known systematics, the review's subject matter—the apparent departure from ΛCDM—is not established, and the 'disentangle new physics from systematics' framing is premature. The review hedges, but the conclusion still elevates a dataset-dependent, calibration-sensitive preference to the status of an indication requiring physical explanation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This review synthesizes the post-DESI literature on the apparent preference for dynamical dark energy (DDE) with w0 > −1, wa < 0, and the possible beyond-ΛCDM interpretations. After summarizing the DESI DR1/DR2 BAO, CMB, and SN constraints, the paper discusses parameterization dependence, the impact of BAO redshift bins and SN calibration, and reconstruction techniques. It then reviews three classes of alternatives to DDE: interacting dark energy (IDE), non-minimally coupled gravity (including hilltop thawing gravity), and non-standard dark matter. A final section considers the implications for the H0, S8, and neutrino-mass tensions. The paper is explicitly hedged in places, but the abstract and conclusion present the DESI-motivated departure from ΛCDM as a robust indication, which is not fully supported by the body's own caveats.","tokens_in":48598,"tokens_out":5191,"duration_ms":56831,"significance":"If the review's central framing is accepted, it provides a timely and fairly comprehensive map of the current DESI-driven model space and usefully emphasizes the degeneracy between DDE, IDE, modified gravity, and non-standard dark matter. Its main strengths are the explicit treatment of parameterization dependence (§2.2), the detailed discussion of SN and BAO systematics (§2.3), and the inclusion of recent recalibrations such as DES-Dovekie. The paper is also well-referenced, though with a heavy concentration of citations to the authors' own program. For a review, the value would be higher if the abstract and conclusion consistently carried the body's caution that the evidence is a dataset-dependent preference rather than an established deviation. The paper does not provide machine-checked proofs or reproducible code, but for a review this is not expected.","major_comments":[{"comment":"The significance of the DDE preference is reported inconsistently. §1 states the range 2.8–3.8σ, §2.1 states 2.5–3.9σ for DR1 combinations and 'up to 4.2σ' for DR2, while §2.3 notes that DES-Dovekie recalibration reduces the DR2 value from about 4.2σ to about 3.2σ. The abstract and conclusion quote only the high end or the uncalibrated value. Please harmonize the numbers and, at the very least, headline the recalibrated range in the abstract and conclusion so that readers are not left with an inflated significance.","section":"§1, §2.1, §2.3 [Eq. (1)]"},{"comment":"The claim that CF and CQ models show 'evidence for non-vanishing dark-sector interactions at the about 3−5σ level' (Li et al. 2026c) is presented as a settled result, but it relies entirely on a single preprint from the authors' own group using their own ePPF/IDECAMB pipeline. The body itself shows that the preference for interaction is model-dependent: §3.1 states that support depends sensitively on the assumed Q form, with only the Q = βH0ρde case giving 'approximately 3σ' (Li et al. 2024b). The conclusion's 'up to 5σ' is therefore misleading. Please report the spread of significances across models and datasets, and explicitly flag the provenance and the lack of independent confirmation.","section":"§3.1, Fig. 4, §5"},{"comment":"Selective reporting in the conclusion: the statement that non-standard dark matter is preferred at 'approximately 2.4σ to 3σ' ignores the paper's own §3.3 finding that, once CPL dark energy is introduced, the significance drops to 0.8–1.1σ (Li et al. 2025e). Likewise, the conclusion's 'IDE up to 5σ' omits the weaker or model-dependent results reported earlier. A review should not cherry-pick the largest significance in the literature; it should give the range and the conditions under which the strongest values arise.","section":"§3.3, §5"},{"comment":"The 'remarkable' Bayes factor of log(B) = 7.34 ± 0.6 for the hilltop thawing gravity model comes from a single analysis (Wolf et al. 2025b). No caveat is given about prior sensitivity, dataset choice, or the fact that this is not an independent cross-check. Given that the review's stated goal is to 'disentangle new physics scenarios from systematic errors,' the Bayesian evidence should be presented with the same caution applied to the DESI-DDE significance. Please add a sentence noting the model- and prior-dependence of such Bayes factors.","section":"§3.2, Eq. (12)–(15), §5"}],"minor_comments":[{"comment":"The significance range is quoted as 2.8–3.8σ in §1 but 2.5–3.9σ in §2.1. Please make these numbers consistent and define which dataset combination each endpoint refers to.","section":"§1, §2.1"},{"comment":"Duplicate reference in the text: 'Yashiki 2025' appears twice in §4.1, and 'Chudaykin et al. 2025' is listed twice in §4.3. The bibliography also contains several entries with the same year but no distinguishing letter. Please clean up the reference list.","section":"§4.1, §4.3"},{"comment":"The final paragraph advertises a future review by the same authors ('Zhang 2026, in preparation'). This is unusual in a journal article and reads as self-promotion. It should be removed or moved to the acknowledgments.","section":"§5"},{"comment":"The paper uses both 'parametrization' and 'parameterization' inconsistently. Pick one spelling for consistency. Also, equations (9) and (10) define w_eff^de and w_eff^c but the notation is not used later; consider either using it or removing it to avoid confusion.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The review is competent and potentially useful, but its central message is systematically tilted toward the authors' own results, especially the IDE claims built on ePPF/IDECAMB and the unified CF/CQ analysis. The abstract and conclusion need to be brought in line with the body's own caveats, particularly the DES-Dovekie recalibration reducing the DDE significance from ~4.2σ to ~3.2σ and the model dependence of the IDE and non-standard-DM significances. I would also flag to the editor the unusual closing reference to a forthcoming review by the same group; this is a presentation issue but also a scope concern for a journal review article."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a review, so judge it as a map. The map is mostly good: it gathers the post-DESI landscape in one place, shows six w(a) parametrizations side by side, lays out the CF/CQ distinction under one code, and frames the real question as \"which ΛCDM assumption is being stressed.\" That framing has practical use. The paper is well hedged about dataset dependence and SN calibration; the discussion of the DESY5 anchor and DES-Dovekie recalibration is honest and current. The authors also flag the LRG z~0.51 bin and the D_V sensitivity. For a review, the literature coverage is broad enough that a newcomer could use it as an entry point.\n\nThe main problem is at the load-bearing joint. The abstract and conclusion present the deviation from ΛCDM as an indication that requires physical explanation. But section 2.3, written by the same authors, says that a few hundredths of a magnitude in SN offset can mimic an evolving EoS, that the DES-Dovekie recalibration lowers the DESY5+DESI significance from ~4.2 to ~3.2σ, and that removing the LRG2 D_V point brings the contour back within 2σ of w=-1. That is not a secure \"indication.\" The stress-test note is right: if these two known systematics drive the trend, the subject matter of the review is not established. The review should either conclude \"dataset-dependent preference, status unresolved\" or separate the framing from the evidence. It hedges in sections, but the framing is too strong.\n\nA second soft spot is asymmetry. The DDE evidence gets extensive caveats; the IDE results, most of which come from the authors' own fits with their own codes, are reported with numbers like 3–5σ and ln B = 7.34 ± 0.6 without equally prominent caveats. The unified CF/CQ analysis is from Li et al. 2026c and the running-coupling reconstruction from Li & Zhang 2025—the group's own analyses of the same DESI data. That does not make them wrong; IDECAMB is public and the machinery is checkable. But a review should apply the same calibration-sensitivity discussion to its preferred alternatives. There are also minor range inconsistencies (2.8–3.8σ in the intro vs 2.5–3.9σ in §2.1) and the conclusion defers the \"unified diagnostic\" to a future review by the same author, which is a self-referential placeholder.\n\nBottom line: this deserves serious peer review as a review article. The organizing value is real, and the honesty of section 2.3 is a credit. But the central framing needs rebalancing and the own-work asymmetry needs fixing. I would cite it as a landscape reference, and I would send it to a referee who knows the DESI likelihoods and SN systematics.","headline":"A competent, genuinely useful DESI-era review whose own section 2.3 undermines the opening claim; caveats about DDE are much sharper than caveats about the authors' own IDE alternatives.","tokens_in":49299,"tokens_out":3115,"would_cite":true,"duration_ms":32691,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"DESI-era data show an apparent preference for dark energy crossing the phantom divide, but the same expansion history can come from interactions, modified gravity, or non-standard dark matter, making the signal an indication, not a discover","keywords":["dark energy","dynamical dark energy","DESI","baryon acoustic oscillations","phantom divide","equation of state","interacting dark energy","cosmological tensions"],"falsifier":"Re-analyze DESI DR2 BAO + Planck CMB + SN data after replacing the DESY5 low-redshift anchor with a recalibrated, homogeneous sample (as in the DES-Dovekie-style updating) and check whether the (w0, wa) contour still excludes (−1, 0) at more than 2σ; separately, remove the z_eff ≈ 0.51 LRG distance point and see whether the contour shifts to consistency with ΛCDM within 2σ, as one cited analysis reports.","tokens_in":47953,"feed_emoji":"🌌","tokens_out":8199,"duration_ms":77506,"temperature":0.7,"pith_summary":"The paper reviews the status of the dark-energy signal suggested by DESI baryon acoustic oscillation data. When combined with CMB and supernova measurements, the data favor a present-day dark energy equation of state w0 > −1 with negative evolution wa < 0, meaning dark energy was more phantom-like in the past and crosses the phantom divide today, at statistical significance between about 2.5σ and 4.2σ depending on the dataset. The review's central organizing claim is that this preference is degenerate: the same distance–redshift relation can be produced by a genuinely evolving dark energy fluid, by energy exchange between dark matter and dark energy, by a scalar field non-minimally coupled to gravity, or by dark matter with a non-zero effective pressure. It concludes that the real question is not whether dark energy evolves, but which assumption of the standard cosmological model is being tested, and how to separate any new physics from supernova calibration and BAO bin effects.","feed_headline":"Dark energy looks changeable in DESI data","feed_subtitle":"BAO, CMB and supernova data favor evolving dark energy at up to 4-sigma—but calibration and competing models blur the picture.","key_machinery":"The central object is the two-parameter CPL form of the dark energy equation of state, w(a) = w0 + wa(1−a), which converts distance measurements into a statement about dark energy's present value and time variation. The review's organizing mechanism is the background-level degeneracy of the expansion history: BAO and supernovae constrain the distance–redshift relation, not the microscopic equation of state, so the same H(z) can be reproduced by an evolving dark energy fluid, a dark-sector interaction (e.g., Q proportional to ρ_de or ρ_c), a scalar-tensor modification of gravity with a non-minimal coupling such as ξφ²R, or a dark matter component with non-zero effective pressure. Breaking thi","core_discovery":"The central claim is that current DESI BAO data, combined with Planck CMB and type Ia supernova compilations, deviate from the cosmological constant in a consistent direction: within the Chevallier-Polarski-Linder type parametrization w(a) = w0 + wa(1−a), the preferred region lies at w0 > −1 and wa < 0, corresponding to an equation of state that was more phantom-like at intermediate redshifts and approaches quintessence today. The statistical significance of this deviation ranges from about 2.5σ to 4.2σ depending on the supernova sample, with the DESY5 compilation giving the strongest preference and PantheonPlus the weakest. The review argues, however, that this background-level departure do","pith_inferences":["A decisive test of the systematics interpretation would be to recompute the DESI+CMB+DESY5 contour using a recalibrated low-redshift supernova anchor; if the preference for w0>−1, wa<0 falls below about 2σ, the signal is a calibration artifact rather than cosmology.","The review's degeneracy argument implies that a growth-rate measurement consistent with ΛCDM would simultaneously disfavor interacting dark energy and non-standard dark matter, leaving evolving dark energy or modified gravity as the remaining explanations — a testable prediction for upcoming redshift surveys.","The neutrino-mass channel offers a cheap discriminator: if future data fix a positive neutrino mass above roughly 0.1 eV within the evolving-equation-of-state model, that supports the phantom-to-quintessence crossing interpretation; if instead the negative-mass anomaly persists, the deviation is more likely a systematic in the matter-power spectrum."],"forward_implications":["If the DESI preference for w0 > −1, wa < 0 persists after supernova recalibration, the cosmological constant alone cannot describe the late-time expansion, and the first credible dynamical-dark-energy signature would be established.","A phantom-divide crossing, if real, rules out single canonical scalar-field dark energy and requires either additional degrees of freedom, interactions, or modified gravity.","The same dataset combination tightens the cosmological neutrino-mass bound to about 0.064 eV in ΛCDM, creating a tension with the normal-hierarchy lower limit; allowing dark energy to evolve relaxes this bound and can turn the neutrino-mass measurement into a positive detection.","The DESI-preferred dark energy does not relieve the H0 tension — it tends to lower the inferred H0 — so a full resolution needs early-universe modifications or interacting dark sectors rather than late-time dynamics alone.","Distinguishing the competing interpretations requires moving from background distances to growth-rate, lensing, and redshift-space-distortion measurements, which carry distinct perturbation-level signatures."],"fun_headline_variants":["DESI data suggest dark energy evolves","Is dark energy constant? DESI data say no","Dark energy may change over cosmic time","DESI finds hints of dynamic dark energy","Evolving dark energy? Evidence from DESI"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the DESI BAO likelihoods, Planck CMB likelihoods, and supernova compilations — especially the heterogeneous low-redshift anchor of DESY5 — encode the true distance–redshift relation closely enough that the inferred w0 > −1, wa < 0 trend reflects cosmology rather than a few-hundredths-of-a-magnitude calibration offset in the supernova sample (or an anomalous BAO bin near z ≈ 0.51).","fun_headline_variants_meta":{"raw":{"variants":["DESI data suggest dark energy evolves","Is dark energy constant? DESI data say no","Dark energy may change over cosmic time","DESI finds hints of dynamic dark energy","Evolving dark energy? Evidence from DESI"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00068,"raw_usage":{"total_tokens":2967,"prompt_tokens":828,"completion_tokens":2139,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":572,"completion_tokens_details":{"reasoning_tokens":2072}},"tokens_in":572,"tokens_out":2139,"duration_ms":15487,"temperature":1.0,"reasoning_tokens":2072,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T10:35:23.175769+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-analyze DESI DR2 BAO + Planck CMB + SN data after replacing the DESY5 low-redshift anchor with a recalibrated, homogeneous sample (as in the DES-Dovekie-style updating) and check whether the (w0, wa) contour still excludes (−1, 0) at more than 2σ; separately, remove the z_eff ≈ 0.51 LRG distance point and see whether the contour shifts to consistency with ΛCDM within 2σ, as one cited analysis reports.","supporting_citations":[],"review_version":2}